This document summarizes an ongoing study examining risk stratification of mild traumatic brain injuries. The study utilizes a prospective observational cohort called HeadSMART that collects clinical data, serum, plasma, mRNA, and DNA samples from patients with head injuries. Advanced imaging and 'omics techniques like proteomics, genomics and metabolomics are used along with clinical data to characterize patients and injuries to improve risk stratification. Preliminary results show certain serum biomarkers like BDNF measured on the day of injury can predict recovery risks. Ongoing work involves examining more biomarkers and developing machine learning models to better prognosticate recovery outcomes after mild traumatic brain injury.
2. Statement of Problem
Korley FK, Pham JC, Kirsch TD: Use of advanced radiology during visits to US emergency
departments
for injury-related conditions, 1998-2007. JAMA 304(13): 1465-71, 2010.
Korley FK, Pham JC, Kirsch TD: Use of advanced radiology during visits to US emergency
departments
for injury-related conditions, 1998-2007. JAMA 304(13): 1465-71, 2010.
4.8 million persons
evaluated in the ED for TBI
each year
2.5 million diagnosed with TBI2.5 million diagnosed with TBI
Korley FK, Kelen GD, Jones CM, Diaz-Arrastia R: Emergency Department Evaluation of
Traumatic Brain Injury in the United States, 2009-2010. J Head Trauma Rehabil. 2015 Sep 10
Korley FK, Pham JC, Kirsch TD: Use of advanced radiology during visits to US emergency
departments for injury-related conditions, 1998-2007. JAMA 304(13): 1465-71, 2010.
3. Hypothesis
A data-driven, multi-disciplinary approach utilizing novel
methods (proteomics, genomics, metabolomics, advanced
imaging) for characterizing patient and injury characteristics,
and coupled with existing clinical data will improve TBI risk-
stratification.
4. Head Injury Serum Markers for Assessing Response to
Trauma (HeadSMART) Cohort
• Prospective observational cohort
• Two demographically distinct academic EDs
• Data: NINDS common data elements
• Serum, plasma and mRNA sampling at 0, 4, 24 hours; 3
and 7 days; 1, 3 and 6 months. DNA at baseline
• Outcome assessment
• Phone
• Battery of cognitive and psychiatric assessments in
person
5. What is TBI? Who should be included in studies?
• American congress of Rehabilitation Medicine’s Definition
• Traumatically induced physiological disruption of
brain function, as manifested by:
• LOC
• Memory loss
• Altered mental status
• Focal neurologic deficit
• What about head injury not meeting “TBI” criteria?
• Head Injury BRain Injury Disputed (HIBRID)
6. Risk of prolonged recovery in HIBRID patients
• To determine the risk of prolonged recovery in HIBRID
patients
• Method:
• Population:
• HeadSMART TBI patients categorized as: HIBRID,
ACRM+ CT-; ACRM+ CT+
• Control groups: Non-head injury trauma controls,
healthy controls
• Outcomes:
• Disability (Glasgow Outcome Scale Extended)
• Post-concussive symptoms (Rivermead Post-
Concussive Questionnaire)
• Depression (Patient Health Questionnaire 9)
8. Patients’ expectations
You were evaluated for a head injury during your visit. What is your
understanding regarding how well you will heal from this head
injury?
Accuracy based
on functional
disability
Accuracy based on
post-concussive
symptoms
Discussed with physicians, high risk (n=7) 57.1% 42.9%
Discussed with physicians, low risk (n=38) 55.3% 60.5%
Did not discuss, high risk (n=9) 100% 75.0%
Did not discuss, low risk (n=38) 60.5% 57.9%
Did not discuss, no idea (n=12) 58.3% (poor),
41.7% (good)
50.0% (poor), 50%
(good)
9. How good is clinician gestalt for identifying high
risk?
Based on what you know now about this patient's presentation, do you think this patient
will have a complete functional recovery i.e. they will be back to their pre-TBI functional
state at 3 months after injury?
Accuracy
based on
functional
disability
Accuracy
based on
having post-
concussive
symptoms
Yes 53.9% 59.4%
No 40.0% 61.6%
Based on what you know now about this patient's presentation, do you think this patient
will have 3 or more post-concussive symptoms (for example: headaches, fatigue, insomnia,
loss of concentration, noise and light sensitivity, memory loss, dizziness) at 3 months after
injury?
Accuracy
based on
functional
disability
Accuracy
based on
having post-
concussive
symptoms
91 – 100%
certain
37.3% 68.9%
71 – 90%
certain
55.6% 52.2%
<70% certain 60.0% 59.5%
12. Ongoing Work
• Examine the diagnostic and prognostic utility of the
following biomarkers in TBI: GFAP, S100B, BDNF,
Troponin, Total tau, phosphorylated Tau, ICAM 5,
Neurogranin, beta synuclein, among others
• Evaluate the effect of catecholamine surge in TBI and its
effect on cerebrovascular reactivity
• Examine the metabolomic profile of recovery from TBI
• Develop prognostic models using machine learning tools
13. Acknowledgements
• Patients and Family Members
• Subject Enrollment
– Hayley Falk M.Sc
– AJ Hall
– Freshta Akabari
– Uju Ofoche
– Olivia Lardo
– Braden Anderson
• Neuropsychiatry
– Alex Vassila B.S.
– Vani Rao M.D.
– Durga Roy M.D.
– Matthew Peters M.D.
– Kostas Lyketsos M.D., M.P.H.
• Neurocognitive/Rehab
– Kathleen Bechtold Ph.D.
• Neurology
– Ramon Diaz-Arrastia M.D., Ph.D
• Proteomics
– Allen Everett, M.D.
– Jenny Van Eyk, Ph.D.
– David Lubman, Ph.D.
• Metabolomics
– Charles Burant, Ph.D.
• Neuroradiology
– Haris Sair M.D.
• Machine learning
– Scott Levin Ph.D.
– Kayvan Najarian, Ph.D.
• Funding
– ImmunArray
– Biodirection
– Robert Wood Johnson Medical
Faculty Development Award
– University of Michigan Injury
Center